THE INHERITANCE OF FLOWER COLOUR IN PISUM 75 



all three crosses are added (see bottom of the table) the deviation of 

 the whites is seen to be almost three .times and the deviation of the 

 purples- even more than three times the standard error. Thus the 

 deviation is really somewhat greater than is generally allowed. 



Table 2. In this table are found (col. 2) all the theoretically con- 

 ceivable factorial combinations in F2 of all the flower colours except 

 the whites, and also (in col. 3 — 7) the theoretical ratio of the segre- 

 gation of their progenies in F3. Col. 9 — 14 show the numerical classi- 

 fication in the respective factorial combinations of the F2 individuals 

 sown both theoretically (col. 10) and observed (col. 9 and 11). In col. 

 8. lastly, reference is given to the tables (3 — 16) in which the numbers 

 obtained in F. from the segregating F. individuals are listed. The 

 observed and the theoretical numbers relating to the numerical sorting 

 of the F2 individuals in the resp. factorial combinations agree parti- 

 cularly well. In most cases the deviation is smaller than the standard 

 error (col. 14). 



The Fo individuals with rose (63 ind.) and light purple coloured 

 flowers (55 ind.) sown in 1912 and found at the bottom of the table 

 do not, however, arise from the crosses I and II in table 1, as in the 

 case of the purple and violet flowered Fo individuals, but owe their 

 origin from a cross 01001 X white which yielded rose flowered Fi and 

 a F2 segregation of rose, light purple, and white in the ratio 9:3:4 

 (table 17). Thus .the constitution of Fi was AaBbcc, and the white 

 parent (from F. arising from a cross, 1907, identical with cross I in 

 table 1) was constituted aaBBcc. 



The fact^that the rose and light purple flowered Fg individuals 

 sown to F3 did not originate from the same cross as the Fo individuals 

 with purple and violet coloured flowers sown to Fv., nor from a nearly 

 like cross, may seem to be an impropriety tending to lessen the value 

 of the investigation. However, this impropriety cannot be of any 

 importance in this case. It must be assumed that the nine rose flowered 

 Fo individuals originating from AaBbcc-Fi-rose are constiuted 

 factorially in exactly the same way as the nine rose flowered F2 in- 

 dividuals from the crosses I and II and from AaBbCc-Fi-purple if the 

 explanation of the inheritance of the flower colours given in the above 

 is correct. And that this is so is evident from the fact that their pro- 

 genies (tables 13 — 15) have showed a segregation exactly the same as 

 the latter nine from crosses I and II theoretically calculated. And what 

 in this case is true of the rose flowered Fo individuals must also be 



